Optimized Data Collection of Relevant Medical Images

1. A computer-implemented method comprising: analyzing a first set of images, including a first image, at an image manipulation stage of an image processing algorithm, wherein the image manipulation stage comprises an image processing algorithmic chain; extracting a plurality of relevant features derived from the analysis at the image manipulation stage, including a first extracted relevant feature with respect to the first image; computing a plurality of degrees of interest, including a first degree of interest, with respect to the plurality of extracted relevant features, wherein a respective degree of interest inversely correlates to a degree of confidence associated with the analysis; determining a subset of the plurality of extracted relevant features from the first set of images wherein a respectively corresponding plurality of degrees of interest exceed a threshold; and in response to determining the subset of the plurality of extracted relevant features from the first set of images, sending the subset of the first set of images to a receiving system.

2. The method of claim 1 , wherein: the image processing algorithmic chain comprises at least one internal operation selected from the group consisting of: detecting a finding in the first image, classifying the finding in the first image, and classifying the first image; and the plurality of degrees of interest are based on a plurality of features derived from an output of the internal operation, wherein the output is selected from the group consisting of: a final output, and an intermediate output.

3. The method of claim 2 wherein: the plurality of features are associated with respectively corresponding measures of confidence with respect to an output of the at least one internal operation; the at least one internal operation implements a model selected from the group consisting of a deep learning model, and a regression model; and the measures of confidence are based on a plurality of values selected from the group consisting of: a distance between the output of the detecting a finding operation, and a given value, a distance between the output of the classifying a finding operation, and a given value, a discrepancy among a plurality of outputs of a respectively corresponding plurality of operations that perform a same task, and a probability of the input data, of an internal operation, to be represented in data used for training the deep learning model corresponding to the internal operation.

4. The method of claim 1 , wherein computing the first degree of interest further comprises: comparing information with respect to the first extracted relevant feature against a second source of information based on a user input; and determining a discrepancy exists between the information with respect to the first extracted relevant feature and the second source of information; wherein: the second source of information comprises information with respect to a plurality of images selected from the group consisting of patient pathologies, finding localizations, and finding characterizations.

5. The method of claim 1 , wherein the image manipulation stage comprises displaying for review the first image, on a system comprising a user interface for user interaction with a displayed image; and the first degree of interest is based on an analysis of a user interaction with respect to the first image.

6. The method of claim 5 , further comprising: generating an analysis result based on an aspect of the user interaction; wherein the aspect of the user interaction is selected from the group consisting of: a time interval duration over which the user interaction takes place, a first user input causing the display device to magnify the image, a second user input causing the display device to shrink the image, a third user input causing the display device to focus on a selected area of the image, a fourth user input causing a second measurement to be taken with respect to an aspect of the first image, and a fifth user input comprising an annotation and associating the annotation with the image.

7. The method of claim 6 , further comprising: pre-localizing a finding of interest with respect to the first image based on the aspect of the user interaction.

8. A computer program product comprising: one or more computer readable storage media, and program instructions collectively stored on the one or more computer readable storage media, the program instructions comprising program instructions programmed to perform: analyzing a first set of images, including a first image, at an image manipulation stage of an image processing algorithm, wherein the image manipulation stage comprises an image processing algorithmic chain; extracting a plurality of relevant features derived from the analysis at the image manipulation stage, including a first extracted relevant feature with respect to the first image; computing a plurality of degrees of interest, including a first degree of interest, with respect to the plurality of extracted relevant features, wherein a respective degree of interest inversely correlates to a degree of confidence associated with the analysis; determining a subset of the plurality of extracted relevant features from the first set of images wherein a respectively corresponding plurality of degrees of interest exceed a threshold; and in response to determining the subset of the plurality of extracted relevant features from the first set of images, sending the subset of the first set of images to a receiving system.

9. The computer program product of claim 8 , wherein: the image processing algorithmic chain comprises at least one internal operation selected from the group consisting of: detecting a finding in the first image, classifying the finding in the first image, and classifying the first image; and the plurality of degrees of interest are based on a plurality of features derived from an output of the internal operation, wherein the output is selected from the group consisting of: a final output, and an intermediate output.

10. The computer program product of claim 9 wherein: the plurality of features are associated with respectively corresponding measures of confidence with respect to an output of the at least one internal operation; the at least one internal operation implements a model selected from the group consisting of a deep learning model, and a regression model; and the measures of confidence are based on a plurality of values selected from the group consisting of: a distance between the output of the detecting a finding operation and a given value, a distance between the output of the classifying a finding operation and a given value, discrepancies between a plurality of outputs of a respectively corresponding plurality of operations that perform a same task, and a probability of the input data, of an internal operation, to be represented in data used for training the deep learning model corresponding to the internal operation.

11. The computer program product of claim 8 , wherein computing the first degree of interest further comprises program instructions programmed to perform: comparing information with respect to the first extracted relevant feature against a second source of information based on a user input; and determining a discrepancy exists between the information with respect to the first extracted relevant feature and the second source of information; wherein: the second source of information comprises information with respect to a plurality of images selected from the group consisting of patient pathologies, finding localizations, and finding characterizations.

12. The computer program product of claim 8 , wherein: the image manipulation stage comprises displaying for review the first image, on a system comprising a user interface for user interaction with a displayed image; and the first degree of interest is based on an analysis of a user interaction with respect to the first image.

13. The computer program product of claim 12 , further comprising program instructions programmed to perform: generating an analysis result based on an aspect of the user interaction; wherein the aspect of the user interaction is selected from the group consisting of: a time interval duration over which the user interaction takes place, a first user input causing the display device to magnify the image, a second user input causing the display device to shrink the image, a third user input causing the display device to focus on a selected area of the image, a fourth user input causing a second measurement to be taken with respect to an aspect of the first image, and a fifth user input comprising an annotation and associating the annotation with the image.

14. The computer program product of claim 13 , further comprising program instructions programmed to perform: pre-localizing a finding of interest with respect to the first image based on the aspect of the user interaction.

15. A computer system comprising: a processor set; and one or more computer readable storage media; wherein: the processor set is structured, located, connected and/or programmed to run program instructions stored on the one or more computer readable storage media; and the program instructions include program instructions programmed to perform: analyzing a first set of images, including a first image, at an image manipulation stage of an image processing algorithm, wherein the image manipulation stage comprises an image processing algorithmic chain; extracting a plurality of relevant features derived from the analysis at the image manipulation stage, including a first extracted relevant feature with respect to the first image; computing a plurality of degrees of interest, including a first degree of interest, with respect to the plurality of extracted relevant features, wherein a respective degree of interest inversely correlates to a degree of confidence associated with the analysis; determining a subset of the plurality of extracted relevant features from the first set of images wherein a respectively corresponding plurality of degrees of interest exceed a threshold; and in response to determining the subset of the plurality of extracted relevant features from the first set of images, sending the subset of the first set of images to a receiving system.

16. The computer system of claim 15 , wherein: the image processing algorithmic chain comprises at least one internal operation selected from the group consisting of: detecting a finding in the first image, classifying the finding in the first image, and classifying the first image; and the plurality of degrees of interest are based on a plurality of features derived from an output of the internal operation, wherein the output is selected from the group consisting of: a final output, and an intermediate output.

17. The computer system of claim 16 wherein: the plurality of features are associated with respectively corresponding measures of confidence with respect to an output of the at least one internal operation; the at least one internal operation implements a model selected from the group consisting of a deep learning model, and a regression model; and the measures of confidence are based on a plurality of values selected from the group consisting of: a distance between the output of the detecting a finding operation and a given value, a distance between the output of the classifying a finding operation and a given value, discrepancies between a plurality of outputs of a respectively corresponding plurality of operations that perform a same task, and a probability of the input data, of an internal operation, to be represented in data used for training the deep learning model corresponding to the internal operation.

18. The computer system of claim 15 , wherein computing the first degree of interest further comprises program instructions programmed to perform: comparing information with respect to the first extracted relevant feature against a second source of information based on a user input; and determining a discrepancy exists between the information with respect to the first extracted relevant feature and the second source of information; wherein: the second source of information comprises information with respect to a plurality of images selected from the group consisting of patient pathologies, finding localizations, and finding characterizations.

19. The computer system of claim 15 , wherein: the image manipulation stage comprises displaying for review the first image, on a system comprising a user interface for user interaction with a displayed image; and the first degree of interest is based on an analysis of a user interaction with respect to the first image.

20. The computer system of claim 19 , further comprising program instructions programmed to perform: generating an analysis result based on an aspect of the user interaction; and pre-localizing a finding of interest with respect to the first image based on the aspect of the user interaction; wherein the aspect of the user interaction is selected from the group consisting of: a time interval duration over which the user interaction takes place, a first user input causing the display device to magnify the image, a second user input causing the display device to shrink the image, a third user input causing the display device to focus on a selected area of the image, a fourth user input causing a second measurement to be taken with respect to an aspect of the first image, and a fifth user input comprising an annotation and associating the annotation with the image.